Hello,
We have some issues with metricbeat to logstash transport. After some time (can be days after startup), metricbeat starts logging "Failed to publish events: unsupported float value: NaN" every report cycle and repeats the same event forever. That leads to billions of copies of one event, that ist delivered to different logstash instances.
Activated metricbesat modules:
- system (cpu, load, core, filesystem, fsstat, memory, network, process, socket)
- docker (container, cpu, diskio, healthcheck, image, info, memory, network)
- elasticsearch (node, node_stats)
- logstash (node, node_stats)
- http (json)
- redis (info, keyspace)
Our logstash transport is encrypted whith TLS.
Versions:
Oracle Linux Server release 7.6
metricbeat 6.5.4 (running in a docker container)
logstash 6.5.4 (running in a docker container)
If we can provide more information to fix this issue let us know.
Thanks and best regards
Sebastian